The recurring slope lineae on Mars have been hypothesized to originate from snow melting, deliquescence, dry flow or shallow groundwater. Except for the dry flow origin, these hypotheses imply the presence of surficial or near-surface volatiles, placing the exploration and characterization of potential habitable environments within the reach of existing technology. Here we present observations from the High Resolution Imaging Science Experiment, heat-flow modelling and terrestrial analogues, which indicate that the source of recurring slope lineae could be natural discharge along geological structures from briny aquifers within the cryosphere, at depths of approximately 750 m. Spatial correlation between recurring slope lineae source regions and multi-scale fractures (such as joints and faults) in the southern mid-latitudes and in Valles Marineris suggests that recurring slope lineae preferably emanate from tectonic and impact-related fractures. We suggest that deep groundwater occasionally surfaces on Mars in present-day conditions.

1. Articles
https://doi.org/10.1038/s41561-019-0327-5
1
Viterbi School of Engineering, University of Southern California, Los Angeles, CA, USA. 2
Department of Geological Applications, National Authority
for Remote Sensing and Space Sciences, Cairo, Egypt. 3
Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA, USA.
*e-mail: heggy@usc.edu
U
nderstanding groundwater flow and its interaction with the
surface is crucial for assessing the potential for recent or
ancient habitability on Mars. The recently discovered recur-
ring slope lineae (RSL)1,2
indicate the presence of seasonal brine
water flow today on the Martian surface3
. RSL emanate from bed-
rock outcrops and progressively lengthen during warm seasons and
fade during cooler seasons preferably on equatorial- and west-facing
slopes1–3
. Slopes (25° to 40°) that encompass active RSL in the south-
ern Martian hemisphere range in temperature from 250 to above
273 K2
, whereas the partial pressure of water vapour at the surface
of Mars is below 1 Pa. It is unlikely that pure liquid water could be
responsible for the relatively long-lasting flow of liquids observed
in RSL locales. Unlike brines, which can sustain a liquid phase at
temperatures lower than 210 K and saturation vapour pressure in
the range of 0.1 to 0.2 Pa4
, pure water can exist only under triple-
point conditions of water temperature (that is, 273.16 K) and vapour
pressure (that is, 611.73 Pa). The origin of RSL has been debated
between dry flows, deliquescence, subsurface melting of brines and
ice and groundwater discharge1,4–9
. Geomorphological analysis and
numerical modelling of geologically recent gullies on Mars, which
are commonly associated with RSL occurrences1
, suggest that these
features are formed either by pure liquid water at high flow rates
(15–60 m3
s−1
) or by briny fluids at much lower rates to produce the
observed short gully channels on Mars10,11
. The incremental length-
ening of RSL throughout most of the year (up to 74% of the Martian
year in Valles Marineris (VM)12
) indicates continuous replenish-
ment and relatively large volumes of water under cold temperature
conditions («273 K) are inconsistent with pure water supply. The
spectral identification of hydrated salts in four RSL locations3
also
supports a briny flow rather than pure water flow.
Except for the dry flow hypothesis, all current hypotheses for
the origin of RSL suggest the presence of a surface or near-surface
source of volatiles that originated these features. Unfortunately,
neither of the radar subsurface sounding experiments (MARSIS
and SHARAD) identified any evidence of shallow groundwater
within the first few hundred metres13,14
in the volcanic terrains in
VM and the southern mid-latitudes, where the majority of RSL
occurrences are reported1,2,12
to support the shallow subsurface
origin of these features. It is worth mentioning that the process-
ing and interpretation of radar sounding data encounter signifi-
cant challenges raised by the resemblance in time delay between
surface clutter and potential subsurface echoes14
. Additionally,
the absence of returned radar echoes from subsurface aquifers
could be attributed to high-conductivity crustal materials that
attenuate the radar signal and obscure the subsurface aquifer13,15
.
Theoretically, orbital sounders with an effective dynamic range
of 24 and 30 dB (equivalent to that of MARSIS and SHARAD)
would achieve a maximum penetration depth ranging from ~300
to 500 m in such volcanic terrain16
.
The recent discovery of a potential 20-km-wide zone of liquid
water at the base of the Martian polar caps at depths exceeding
1.5 km below the surface17
draws attention to the presence of deep
aquifers on Mars in present-day conditions. Hence, alternative for-
mation mechanisms associated with deep groundwater dynamics
need to be explored, as the source of recharge of RSL remains largely
unconstrained and poorly comparable with appropriate Earth ana-
logues. The term ‘deep’ groundwater in this study is assigned to
potential aquifers that lie beneath the maximum theoretical pen-
etration depth (~500 m) of orbital sounding radar sensors in the
Martian volcanic terrains, and hence remained uncovered and
required a geological interpretation rather than geophysical prop-
ping to demonstrate their existence.
Here, we investigate the possibility of a deep groundwater ori-
gin of the RSL by conducting a structural mapping of RSL source
regions along the walls of three craters in the southern mid-lati-
tudes and in VM. We describe the spatial correlation between tec-
tonic and impact-related faults and RSL sources and we then use a
heat-flow model and terrestrial analogues to constrain the ambi-
guities associated with the depth of RSL recharge. In what follows,
we examine whether RSL are structurally controlled artesian dis-
charge from deep sources rather than near-surface or surficial ones,
as widely hypothesized.
A deep groundwater origin for recurring slope
lineae on Mars
Abotalib Z. Abotalib 1,2
and Essam Heggy 1,3
*
The recurring slope lineae on Mars have been hypothesized to originate from snow melting, deliquescence, dry flow or shallow
groundwater. Except for the dry flow origin, these hypotheses imply the presence of surficial or near-surface volatiles, placing
the exploration and characterization of potential habitable environments within the reach of existing technology. Here we pres-
ent observations from the High Resolution Imaging Science Experiment, heat-flow modelling and terrestrial analogues, which
indicate that the source of recurring slope lineae could be natural discharge along geological structures from briny aquifers
within the cryosphere, at depths of approximately 750 m. Spatial correlation between recurring slope lineae source regions and
multi-scale fractures (such as joints and faults) in the southern mid-latitudes and in Valles Marineris suggests that recurring
slope lineae preferably emanate from tectonic and impact-related fractures. We suggest that deep groundwater occasionally
surfaces on Mars in present-day conditions.
Nature Geoscience | VOL 12 | APRIL 2019 | 235–241 | www.nature.com/naturegeoscience 235

2. Articles Nature Geoscience
Correlating RSL occurrences to structural features
Confirmed RSL in Palikir Crater are reported as dark and low-
albedo streaks with 0.5 to 5 m width, and up to hundreds of metres
long, which lasted three Martian years (that is, Martian years 28, 29
and 30)1
. On the basis of multi-temporal High Resolution Imaging
ScienceExperiment(HiRISE)images,weextendtheunceasingrecur-
ring behaviour into the 33rd Martian year (Supplementary Fig. 1).
We also report RSL activity in two other craters in the southern mid-
latitudes, including an unnamed crater to the south of Palikir Crater
inside Newton Crater Basin and Triolet Crater. All of the reported
locations show RSL sources from highly deformed bedrocks (Fig. 1).
General inspection of the HiRISE images indicates that the craters
have impact-related fractures remarkably similar to those reported
from terrestrial craters (for example, Lonar and Meteor craters;
see refs. 18,19
). The rims of these craters are dissected by concentric
and radial fractures that can be seen from the plan view (Fig. 1). We
map impact-related multi-scale fractures (that is, joints and faults) in
the three craters using five main characteristics as described in ref. 20
,
which include: displacement of beds along linear planes; subparallel
alignment of topography; presence of scarps; triangular facets; and
linear streams (Supplementary Fig. 2).
Concentric fractures strike parallel to the impact rims and
extend downslope. These fractures appear as resistant ridges along
gullies and RSL-related channels. On the other hand, the radial
fractures are generally vertical on the crater wall and intersect the
concentric fractures at multiple locations. We note from the HiRISE
images that the majority of RSL onsets emanate from concentric
fractures where they intersect radial fractures along the free faces
of the crater walls (Fig. 1). In Triolet Crater, RSL source areas are
located along the intersection of concentric fractures and the pre-
impact radial grabens associated with the rise of Tharsis21
(Fig. 1).
Along fractured zones in the studied craters, varied numbers of RSL
(that is, range from 1 to more than 30 streaks) initiate and extend
downslope for hundreds of metres before they terminate.
VM, where more than half of the RSL locations on Mars are
reported12
, is a fault-controlled canyon that extends for 4,000 km
and is associated with extensional tectonics22
leading to one of
the largest normal fault systems in the Solar System (Fig. 2). The
troughs of VM contain a series of erosion-resistant ridges that
stretch over tens of kilometres and are known as fault trace ridges
with a characteristic dog-leg offset23
. These ridges are attributed to
fault zone cementation by ascending groundwater circulations23
.
Superimposing the distribution of normal faults along the walls
of VM, fault trace ridges and the confirmed and candidate occur-
rences of RSL on a Mars Orbiter Laser Altimeter (MOLA) and
High-Resolution Stereo Camera (HRSC) blended digital eleva-
tion model (Fig. 2a) indicates a strong spatial correlation between
the locations of RSL and the faulted areas in VM. A close-up view
using HiRISE imagery of an RSL occurrence in Coprates Chasma
(Fig. 2b,c), where the highest areal density of RSL on Mars is
reported, indicates that RSL emanate from a densely faulted zone
near the top of a fault trace ridge. However, RSL emanate from a
dusty-covered side that impedes the demonstration of the fault con-
trol in the RSL source region, yet the adjacent side of the ridge shows
b
c
a
Concentric fractures
Radial fractures
d
N
m
0
N
m
m
0
0 100
800
200 m
N
m
0 50100
Fig. 1 | RSL locations along fractured crater walls in the southern mid-latitudes of Mars. a, Palikir Crater ESP_022689_1380. b, An unnamed crater within
Newton Crater Basin ESP_040491_1375. c, Triolet Crater ESP_022808_1425. The blue arrows refer to concentric fractures and the red arrows refer to
the pre-impact radial grabens. The inset shows a close-up view of the intersection area between the concentric fractures and the pre-impact grabens
and is highlighted in c with a purple box. d, A schematic diagram of crater wall fractures showing the distribution of radial and concentric fractures within
terrestrial basalts and sandstones. Credit: Panel d adapted with permission from ref. 18
, Wiley.
Nature Geoscience | VOL 12 | APRIL 2019 | 235–241 | www.nature.com/naturegeoscience236

3. ArticlesNature Geoscience
a clear exposure of the same geological units from which RSL ema-
nate (Fig. 20B). These geological units show a dense deformation
and multiple displacements of bedding planes including the dog-
leg offset (Fig. 20C) that characterizes the fault trace ridges in VM23
.
Evidence of artesian groundwater discharge is commonly reported
near the locations of RSL in VM by using geomorphological and
geological analyses24
and experimental and numerical modelling25
.
Our observations from RSL locations demonstrate a high struc-
tural control on the emergence of these features. Such control
implies a discharge of subsurface water/brines in the liquid phase
from bedrock exposures and opposes previous hypotheses for the
formation of RSL by deliquescence7
, slope destabilization by meta-
stable boiling ice melts8
or granular flow9
. Nevertheless, the hypoth-
esis of deep briny groundwater accessing the Martian surface as
springs and seeps was challenged by the widespread occurrences of
RSL along crater walls, ridges and central peaks in complex craters1
,
rather than along the lowlands inside crater basins where ground-
water discharge from regional flow in subsurface aquifers is most
likely to occur. Yet, the structural control on the discharge of these
subsurface brines, as described in this study, can provide a plausible
explanation for this phenomenon.
Palikir Crater as a case study
Theperfectlyexposedbedrocksandminimaldustcoveralongthecra-
ter walls, availability of numerous multi-temporal HiRISE and surface
temperature images and a high-resolution (up to 1 m) digital terrain
model (HiRISE DTM), in addition to the well-established recurrence
of RSL activity for the last six Martian years, all make Palikir Crater
an optimal site to further examine the hypothesis of a deep ground-
water source of RSL along vertical faults and fractures. We derive a
20-m interval contour map from the HiRISE DTM over Palikir Crater
to determine the elevations of RSL onsets and concentric fractures.
We consider fractures that intersect more than two contour lines as
radial fractures. Superimposing the contour map over the orthogonal
HiRISE image indicates that RSL onsets occur at different elevations
(Fig. 3) ranging from −740 to 0 m with two peaks of preferential dis-
charge at elevations from −200 to −280 m and at −360 m (Fig. 4a).
We observe that concentric fractures are also abundant at the same
elevations where RSL onsets are abundant and the frequencies of both
RSL onsets and concentric fractures decrease significantly at lower
elevations away from the crater rim. This decrease in frequency could
be attributed either to the burial of rock exposures under the cra-
ter fill materials26
or to the natural decrease of fault activity towards
the crater centre27
. Moreover, the Pearson correlation coefficient
(PCC; Fig. 4b) between the frequencies of concentric fractures and
RSL onsets shows a significant positive correlation between the two
datasets (P ≈ 0.67).
Most importantly, RSL onsets originate at discrete and different
elevations along the same vertical path downslope from the crater
rim (Fig. 3). It is most likely that, at specific locations, when a set of
concentric fractures at different elevations intersect the same radial
fracture, RSL onsets initiate at each point of intersection, where the
intensified structural deformation along the intersection between
different fault sets provides a preferential pathway for subsurface
liquid to surface on Mars.
Given the elevation range of mapped fractures along the crater
walls in Palikir Crater (that is, −740 to 0 m), and if we consider that
RSL represent discharge zones of groundwater along fractures, a
localized aquifer that is deeper than the above-mentioned elevation
range beneath the crater should exist and groundwater should move
upward from this aquifer and locally discharge along fractures.
Vertical groundwater flow and artesian discharge on Mars
The transient upward flow of a briny aquifer (q) is given by
a
b
c
Dog-leg offset
Multiple displacements
Confirmed and candidate RSL
Normal faults
Major fault trace ridges
Elevation value (m)
High: 6,844
Low: –5,472
Tithoniun
RSLIus
Melas
Ophir
Coprates
0 500
0
0 200
m 300
m
km
Fig. 2 | RSL occurrences in VM. a, The distribution of normal faults22
, fault trace ridges23
and RSL occurrences12
over a MOLA/HRSC mosaic of VM
showing the spatial correlation between these features. b, The emergence of RSL in Coprates Chasma along a highly deformed zone (yellow dashed
lines) (HiRISE image ESP_038285_1665). The location of b is located in a as a red star. c, Detail from the area outlined in orange in b showing multiple
displacements of a marker horizontal stratum along faults (white arrows), and a dog-leg offset along a linear ridge (yellow arrows), indicating high
deformation of RSL source regions.
Nature Geoscience | VOL 12 | APRIL 2019 | 235–241 | www.nature.com/naturegeoscience 237

4. Articles Nature Geoscience
= −
∂
∂
q K
h
t
(1)
where the hydraulic conductivity (K) and the negative hydrau-
lic gradient (∂
∂
h
t
) control the artesian upward flow of these brines.
The briny aquifer is expected to be under positive pressure that has
been exerted by the weight of the overlying impermeable crystalline
rocks and the cryosphere. Although earlier hydrological models of
Mars28
indicate that the cryosphere must be thick enough (>1 km) to
confine an aquifer, recent models such as the cryovolcanic model29
indicate that changes in surface temperature cause impingement of
the cryosphere or freezing front onto liquid water aquifers confined
within the cryosphere at depths of several hundred metres. Freezing
of the aquifers increases pore pressures and allows buoyant water
to rapidly ascend toward the surface. Moreover, artesian flow from
confined aquifers several hundred metres below the surface of Mars
was also modelled in refs. 10,30,31
.
The hydraulic conductivity of fault zones in crystalline rocks
is several orders of magnitude higher than in the host rocks (for
example, 10−6
m s−1
along fault zones compared to 10−13
m s−1
within
the crystalline rocks in the high Arctic (see ref. 32
)). As the cra-
ter walls and central peaks represent areas with intensive faulting
and impact-related deformation27
, fracture zones along faults will
focus groundwater and artesian discharge of brines, and associated
RSL formation will preferentially occur along these zones (Fig. 5).
Similar observations were reported from the Yucatan Peninsula33
where significant deep circulating groundwater preferentially dis-
charges to the surface in correlation with the peak ring and cra-
ter rims of the buried Chicxulub Crater. Similarly, in Lonar Crater,
India, which is formed in the Deccan Trap basalts, active springs
occur at different elevations (for example, 574, 550 and 510 m) that
are consistent with highly deformed and conductive horizons34
. The
phenomenon of preferential discharge of deep groundwater under
high hydrostatic pressures along fault-related ridges and scarps is
also well documented in the Sahara35
. The conceptual model (Fig. 5)
also suggests that seasonal melting and freezing of the shallow
subsurface acts as a gauge controlling the RSL activity. The system
shuts down during winter seasons, when the ascending near-surface
water freezes within fault pathways, and resumes during summer
seasons when brine temperatures rise above the freezing point.
Concentric fractures Radial fractures RSL onset
a b
–40
–100
–160
1000
m
1000
m
–180
–160
–220–260
Fig. 3 | Fault control on RSL emergence in Palikir Crater. a,b, The intersection of concentric and radial fractures along the wall of Palikir Crater during
winter seasons ESP_021555_1380 (a) and during summer seasons ESP_022689_1380 (b). Note the emergence of RSL during the summer season along
discrete elevations in concordance with the locations of fractures. RSL are indicated with black arrows and elevation values are in purple.
b
Concentric faults
RSL onset
a
–80 to –99 –280 to –299 –480 to –499 –680 to –699
Elevation (m)
Frequency
0
10
20
30
40
0 10 20 30 40
Fault frequency
10
0
20
30
40
RSLfrequency
PCC = 0.6675
Fig. 4 | Correlation between faults and RSL. a, The frequencies of
concentric faults and RSL onsets showing a strong correlation at
different elevations along the crater walls. b, The PCC between the
frequencies of RSL and concentric faults showing a positive correlation
between the two datasets. The frequencies of RSL onsets and fractures
are counted in fixed 20 m buffer zones and elevation values are
obtained from a 20 m contour map; hence, all of the measurements
have an uncertainty of 20 m. The trendline and the line of
perfect correspondence are presented as a red line and a black
dashed line, respectively.
Nature Geoscience | VOL 12 | APRIL 2019 | 235–241 | www.nature.com/naturegeoscience238

5. ArticlesNature Geoscience
Heat-flow modelling of vertical groundwater flow
To evaluate whether deep brine groundwater discharge into the
surface along impact-related fractures in Palikir Crater is consistent
with our observations, and to constrain the depth to the aquifer, we
implement the combined heat-flow model developed in ref. 36
. The
model is described in detail in the Methods.
The model results (Fig. 6) show that the flow parameters (that
is, the values of ϒ) do not affect the outflow temperatures of the
discharge under present-day geothermal gradient of Mars. On
the other hand, the depth of the aquifer has the greatest effect on
the outflow temperature. The outflow temperature during sum-
mer seasons reaches 343 and 312 K for aquifer depths of ~4.5 km
and ~750 m, respectively, and α is given as 20 °C km−1
. Using α of
15 °C km−1
, the outflow temperatures are 333 and 310 K for aqui-
fer depths of ~4.5 km and ~750 m, respectively. The outflow tem-
peratures of the previous case are higher than the freezing point of
liquid brines in the subsurface (that is, <240 K; see ref. 5
). On the
other hand, the outflow temperature during winter seasons declines
to 270 and 233 K and to 255 and 231 K for the same depths at α of
20 and 15 °C km−1
, respectively. The predicted outflow temperature
during winter seasons for an aquifer depth of ~750 m beneath RSL
locales is slightly lower than the freezing point of brines ( < 240 K;
see ref. 5
). On the other hand, a deeper aquifer (for example, ~4.5 km
beneath the Martian cryosphere) would produce outflow tempera-
tures that are higher than the freezing point of brines and thus a
continuous discharge of RSL should be expected. This scenario is
not consistent with the reported seasonality and recurring charac-
teristics of RSL. Therefore, a natural discharge of briny aquifer at
depths of about ~750 m below the surface along geological struc-
tures is more consistent with our observations in the RSL locations
and hence it could provide a likely potential source of recharge
for RSL in Palikir Crater and in Martian southern mid-latitudes.
Subsurface intra-permafrost taliks within the Martian cryosphere
could provide a recharge source for the deep brines as well as for
the case of continental Antarctica37
. The modelled seasonal melt-
ing and freezing of the groundwater discharge is consistent with a
finding from hydrological modelling38
, which indicates that shallow
unconfined and confined semi-pervious (for example, hydraulic
conductivity of 10−6
m s−1
) briny aquifers will freeze under winter
mid and high southern latitudes and discharge restarts in the sum-
mer favouring equator-facing slopes. All reported RSL source areas
on Mars in VM2,12
, northern mid-latitudes6
and southern mid-lati-
tudes1,39
are found on the slopes of crater walls and faulted canyons.
It is not coincidental that, despite their scattered distribution, RSL
are restricted to highly fractured and deformed zones. The spatial
limitation of RSL to highly deformed areas indicates, by default, a
RSL development
T > 0
K =1 0–13
m s–1
T < 0
K = 10–13
m s–1
Faults
Intra-permafrost talik
Unfrozen active layer Ejecta Permafrost (volcanic rocks)
Crater-fill deposits Artesian upward leakage
Crater
200m
a
b
K=10–6
m
s–1
K
=
10–6
m
s–1
Fig. 5 | The control of seasonal melting and freezing of the shallow subsurface on RSL activity. a, During winter seasons the system shuts down when
ascending brines freeze within fault pathways in the near-surface. b, During summer seasons the system resumes when the brine temperature rises above
the freezing point.
Nature Geoscience | VOL 12 | APRIL 2019 | 235–241 | www.nature.com/naturegeoscience 239

6. Articles Nature Geoscience
global control of geological structures on the formation of RSL on
Mars. However, the demonstrated role of deep groundwater dis-
charge along faults and fractures in the formation of RSL does not
rule out other mechanisms of formation of RSL that could be differ-
ent for specific areas. The lack of global coverage of HiRISE imagery,
the widespread geographic distribution of RSL1,2,6
and their occur-
rence among diverse geologic settings and broad elevation ranges40
make a sole mechanism for the formation of RSL out of reach.
Online content
Any methods, additional references, Nature Research reporting
summaries, source data, statements of data availability and asso-
ciated accession codes are available at https://doi.org/10.1038/
s41561-019-0327-5.
Received: 10 September 2018; Accepted: 13 February 2019;
Published online: 28 March 2019
References
1. McEwen, A. S. et al. Seasonal flows on warm Martian slopes. Science 333,
740–743 (2011).
2. McEwen, A. S. et al. Recurring slope lineae in equatorial regions of Mars.
Nat. Geosci. 7, 53–58 (2014).
3. Ojha, L. et al. Spectral evidence for hydrated salts in recurring slope lineae
on Mars. Nat. Geosci. 8, 829–832 (2015).
4. Martinez, G. M. & Renno, N. O. Water and brines on Mars: current evidence
and implications for MSL. Space Sci. Rev. 175, 29–51 (2013).
5. Chevrier, V. F. & Rivera-Valentin, E. G. Formation of recurring slope lineae
by liquid brines on present-day Mars. Geophys. Res. Lett. 39, L21202 (2012).
6. Stillman, D. E., Michaels, T. I., Grimm, R. E. & Hanley, J. Observations
and modeling of northern mid-latitude recurring slope lineae (RSL)
suggest recharge by a present-day Martian briny aquifer. Icarus 265,
125–138 (2016).
7. Heinz, J., Schulze‐Makuch, D. & Kounaves, S. P. Deliquescence‐induced
wetting and RSL‐like darkening of a Mars analogue soil containing various
perchlorate and chloride salts. Geophys. Res. Lett. 43, 4880–4884 (2016).
8. Massé, M. et al. Transport processes induced by metastable boiling water
under Martian surface conditions. Nat. Geosci. 9, 425–428 (2016).
9. Schmidt, F., Andrieu, F., Costard, F., Kocifaj, M. & Meresescu, A. G.
Formation of recurring slope lineae on Mars by rarefied gas-triggered
granular flows. Nat. Geosci. 10, 270–273 (2017).
10. Heldmann, J. L. & Mellon, M. T. Observations of Martian gullies and
constraints on potential formation mechanisms. Icarus 168, 285–304 (2004).
11. Heldmann, J. L. et al. Formation of Martian gullies by the action of liquid
water flowing under current Martian environmental conditions.
J. Geophys. Res. 110, E05004 (2005).
12. Stillman, D. E., Michaels, T. I. & Grimm, R. E. Characteristics of the
numerous and widespread recurring slope lineae (RSL) in Valles Marineris,
Mars. Icarus 285, 195–210 (2017).
13. Farrell, W. M. et al. Is the Martian water table hidden from radar view?
Geophys. Res. Lett. 36, L15206 (2009).
14. Nunes, D. C. et al. Examination of gully sites on Mars with the shallow radar.
J. Geophys. Res. 115, E10004 (2010).
15. Heggy, E. et al. On water detection in the martian subsurface using
sounding radar. Icarus 154, 244–257 (2001).
16. Heggy, E. et al. Ground penetrating radar sounding in mafic lava flows:
Assessing attenuation and scattering losses in Mars analog volcanic terrains.
J. Geophys. Res. 111, E06S04 (2006).
17. Orosei, R. et al. Radar evidence of subglacial liquid water on Mars.
Science 361, 490–493 (2018).
18. Kumar, P. S. & Kring, D. A. Impact fracturing and structural modification
of sedimentary rocks at Meteor Crater, Arizona. J. Geophys. Res. 113,
E09009 (2008).
19. Kumar, P. S., Head, J. W. & Kring, D. A. Erosional modification and gully
formation at Meteor Crater, Arizona: insights into crater degradation
processes on Mars. Icarus 208, 608–620 (2010).
20. Singhal, B. B. S. & Gupta, R. P. in Applied Hydrogeology of Fractured Rocks
(Springer Science & Business Media, 2010).
21. Andrews‐Hanna, J. C., Zuber, M. T. & Hauck, S. A. Strike‐slip faults on
Mars: observations and implications for global tectonics and geodynamics.
J. Geophys. Res. 113, E08002 (2008).
22. Montgomery, D. R. et al. Continental-scale salt tectonics on Mars and
the origin of Valles Marineris and associated outflow channels.
Geol. Soc. Am. Bull. 121, 117–133 (2009).
23. Treiman, A. H. Ancient groundwater flow in the Valles Marineris on Mars
inferred from fault trace ridges. Nat. Geosci. 1, 181–183 (2008).
24. Montgomery, D. R. & Gillespie, A. Formation of Martian outflow channels by
catastrophic dewatering of evaporite deposits. Geology 33, 625–628 (2005).
25. Marra, W. A., Braat, L., Baar, A. W. & Kleinhans, M. G. Valley formation by
groundwater seepage, pressurized groundwater outbursts and crater-lake
overflow in flume experiments with implications for Mars. Icarus 232,
97–117 (2014).
26. Osinski, G. R. & Lee, P. Intra‐crater sedimentary deposits at the Haughton
impact structure, Devon Island, Canadian High Arctic. Meteorit. Planet. Sci. 40,
1887–1899 (2005).
27. Osinski, G. R. & Spray, J. G. Tectonics of complex crater formation as
revealed by the Haughton impact structure, Devon Island, Canadian
High Arctic. Meteorit. Planet. Sci. 40, 1813–1834 (2005).
28. Carr, M. H. Formation of Martian flood features by release of water from
confined aquifers. J. Geophys. Res. 84, 2995–3007 (1979).
29. Gaidos, E. J. Cryovolcanism and the recent flow of liquid water on Mars.
Icarus 153, 218–223 (2001).
30. Mellon, M. T. & Phillips, R. J. Recent gullies on Mars and the source of liquid
water. J. Geophys. Res. 106, 23165–23179 (2001).
31. Hauber, E. et al. Asynchronous formation of Hesperian and Amazonian
aged deltas on Mars and implications for climate. J. Geophys. Res. 118,
1529–1544 (2013).
32. Scheidegger, J. M., Bense, V. F. & Grasby, S. E. Transient nature of Arctic
spring systems driven by subglacial meltwater. Geophys. Res. Lett. 39,
L12405 (2012).
33. Pope, K. O., Rejmankova, E. & Paris, J. F. Spaceborne imaging radar-C
(SIR-C) observations of groundwater discharge and wetlands associated
with the Chicxulub impact crater, northwestern Yucatan Peninsula, Mexico.
Geol. Soc. Am. Bull. 113, 403–416 (2001).
34. Komatsu, G. et al. Drainage systems of Lonar Crater, India: contributions
to Lonar Lake hydrology and crater degradation. Planet. Space Sci. 95,
45–55 (2014).
35. Abotalib, A. Z., Sultan, M. & Elkadiri, R. Groundwater processes in Saharan
Africa: implications for landscape evolution in arid environments.
Earth Sci. Rev. 156, 108–136 (2016).
36. Andersen, D. T., Pollard, W. H., McKay, C. P. & Heldmann, J. Cold springs in
permafrost on Earth and Mars. J. Geophys. Res. 107, 5015 (2002).
0 2 4 6 8 10 12 14
213
233
253
273
293
313
333
353
a
b
343
312
268
229
α = 20 °C
α = 15 °C333
310
257
228
Freezing point
Freezing point
213
233
253
273
293
313
333
353
Tw(K)Tw(K)
Υ (km)
Fig. 6 | Modelled outflow temperatures of groundwater discharge along
the surface of Palikir Crater fractured walls. The outflow temperature (Tw)
is modelled as a function of the flow parameter (ϒ) during winter (black
lines) and summer (red lines) seasons under aquifer depths (zb) of 750 m
(solid lines) and 4.5 km (dashed lines). a,b, The outflow temperatures
at a geothermal gradient of 20 and 15 °C km−1
, respectively. ϒ at values
of 2.79 km and 3.21 km is shown as solid blue and dashed blue lines for
aquifer depths of 750 m and 4.5 km, respectively.
Nature Geoscience | VOL 12 | APRIL 2019 | 235–241 | www.nature.com/naturegeoscience240

7. ArticlesNature Geoscience
37. Forte, E., Dalle Fratte, M., Azzaro, M. & Guglielmin, M. Pressurized brines
in continental Antarctica as a possible analogue of Mars. Sci. Rep. 6,
33158 (2016).
38. Goldspiel, J. M. & Squyres, S. W. Groundwater discharge and gully formation
on martian slopes. Icarus 211, 238–258 (2011).
39. Stillman, D. E., Michaels, T. I., Grimm, R. E. & Harrison, K. P. New
observations of Martian southern mid-latitude recurring slope lineae (RSL)
imply formation by freshwater subsurface flows. Icarus 233, 328–341 (2014).
40. Chojnacki, M. et al. Geologic context of recurring slope lineae in Melas and
Coprates Chasmata, Mars. J. Geophys. Res. 121, 1204–1231 (2016).
Acknowledgements
The authors are grateful to M. Sultan from Western Michigan University, R. Elkadiri
from Middle Tennessee State University, H. El Safty from USC and Y. Gim from JPL for
the discussions that helped to generate this manuscript. The first author is a postdoctoral
research associate currently funded by the University of Southern California under
the NASA Planetary Geology and Geophysics award NNX15AV76G awarded to the
principal investigator E.H.
Author contributions
A.Z.A. and E.H. designed the project, A.Z.A. performed the measurements, and A.Z.A.
and E.H. wrote the manuscript.
Competing interests
The authors declare no competing interests.
Additional information
Supplementary information is available for this paper at https://doi.org/10.1038/
s41561-019-0327-5.
Reprints and permissions information is available at www.nature.com/reprints.
Correspondence and requests for materials should be addressed to E.H.
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations.
© The Author(s), under exclusive licence to Springer Nature Limited 2019
Nature Geoscience | VOL 12 | APRIL 2019 | 235–241 | www.nature.com/naturegeoscience 241

8. Articles Nature Geoscience
Methods
For our investigation of RSL activity and fracture mapping along the crater
walls of the three studied locations (that is, Palikir Crater, Triolet Crater and
an unnamed crater to the south of Palikir Crater within the Newton Crater
Basin) and in VM, we use multiple orthorectified and Reduced Data Record
(RDR) HiRISE images. For Palikir Crater, we use only orthorectified images
(ESP_021555_1380, ESP_030614_1380 and ESP_022689_1380) to ensure
coalignment with the topographic datasets derived from the high-resolution
DTM (DTEEC_005943_1380_011428_1380) at ~1 m pixel resolution and an
estimated vertical precision of ~30 cm41
. The DTM is downloaded from the Lunar
and Planetary Laboratory at the University of Arizona website (https://www.
uahirise.org/PDS/DTM/PSP). For Triolet Crater and the other unnamed crater
and for VM, we use RDR HiRISE images that show the recurring behaviour of the
RSL activity in these locations including ESP_022808_1425, ESP_023586_1425,
ESP_025168_1375, ESP_040491_1375, ESP_038285_1665, ESP_040171_1665.
For the studied locations, both the 6-km-wide HiRISE visible-wavelength (RED)
band-pass images and the 1-km-wide enhanced-colour (infrared–red–blue/
green or IRB) images are used to facilitate the investigation of RSL distribution
and fault mapping. RDR HiRISE images are georeferenced and overlaid using
ArcGIS version 10.5 to ensure the orthorectification of multiple images covering
the same area. For regional spatial analysis in VM, we use a mosaic of blended
digital elevation model data derived from the MOLA and the HRSC. The mosaic
is downloaded from the USGS Astrogeology Science Center website (https://
astrogeology.usgs.gov/search/map/Mars/Topography/HRSC_MOLA_Blend/
Mars_HRSC_MOLA_BlendDEM_Global_200mp_v2). We develop a systematic
mapping approach of impact-related multi-scale fractures (joints and faults)
using the procedures described in ref. 20
within a geographic information system
environment. We derive a 20-m interval contour map from the HiRISE DTM
over Palikir Crater to determine the elevations of RSL onsets and concentric
fractures and consider fractures that intersect more than two contour lines as radial
fractures. We also consider a buffer zone of 20 m around the mapped RSL onsets to
avoid user errors associated with visual interpretation of the initiation of RSL along
the slopes of the crater walls. Hence, all measured elevation values of RSL and faults
have an uncertainty of 20 m. We create point and polyline shapefiles to map the
location of RSL onsets and fractures, respectively, where each shapefile includes the
elevation values of each feature that are extracted from the DTM. The frequency
of RSL onsets and fractures at elevation intervals of 20 m and the PCC between
the two datasets are calculated in MATLAB R2013b. The PCC is a measure of the
correlation between two variables following equation (2):
=
∑ − − ̄
∑ − ∑ − ̄
=
= =
X X Y Y
X X Y Y
PCC
( ) ( )
( ) ( )
(2)
i
n
i i
i
n
i i
n
i
1
1
2
1
2
where X and Y are the two variables, and X and ̄Y are the variable means.
We calculated the PCC between fault frequency and RSL frequency and we
obtained a value of 0.67 indicating a significant positive correlation between fault
presence and RSL detection.
We also use a combined heat-flow model36
to simulate the outflow temperature
of groundwater discharge along the walls of Palikir Crater using MATLAB R2013b.
Description of the heat-flow model. The model was originally constructed to
simulate the outflow temperature of groundwater discharge from different aquifer
depths along faults cutting through the permafrost of Earth as an analogue to Mars
and to demonstrate that liquid water can reach the surface in regions of thick and
continuous permafrost through a vertical conductive structure in the absence of
volcanic heat sources. The Arctic springs that are studied in ref. 36
show the exact
temperature-dependence behaviour that we report for the RSL, where springs are
active when they discharge underneath wet-based parts of the ice and the system
freezes when it is exposed to the surface temperature with thinning ice33,36
.
Assuming that the impact-related faults in Palikir Crater, along which
ascending brines occur, are cylinders with symmetric temperature profiles, and
considering time-averaged conditions, the mean temperature of the outflow of
brines is given by
αγ α γ= − + +γ
−
T z T z( ) e ( ) (3)
z z
w 0
b
where Tw is the temperature of the liquid brines as they flows upward along the
faults at depth z; T0 is the average seasonal surface temperature; α is the Martian
geothermal gradient; zb is the depth to the briny aquifer; and γ is a characteristic
scale length for the spring outflow considering every RSL source area as a spring,
which is calculated as
γ =
̇
π
∕∞
mc
k
r r
2
ln( ) (4)0
where ̄Y is the mass flow rate, c is the specific heat of liquid brines, k is the thermal
conductivity, r0 is the radius of the spring and r∞ can be considered the distance
between the RSL elevation and the depth to the aquifer36
.
The values used for T0 were 298 and 220 K, which represent the average near-
maximum surface temperature during summer and winter seasons, respectively39
.
The values used for the geothermal gradient α were 20 °C km−1
(ref. 42
) and
15 °C km−1
(ref. 43
) and the values used for the depth to the aquifer zb were 4.5 km
and 750 m to represent deep conditions (that is, beneath the Martian cryosphere)
or relatively shallower conditions such as a layer of year-round unfrozen ground
(that is, taliks) inside the permafrost. The geothermal gradient provides the heat
source of the deep groundwater at the designated aquifer depths (that is, 4.5 km
and 750 m). The depth value of 4.5 km corresponds to the expected aquifer depths
beneath the cryosphere in the southern mid-latitudes42,43
and the depth value of
750 m is consistent with the RSL observations in Palikir Crater.
The values of γ were calculated as 3.21 km and 2.79 km for aquifer depths of
4.5 km and 750 m, respectively. These values are obtained by calculating the mass
flow rate ( ̇m) from the hydromorphic characteristics of RSL in Palikir Crater as
described in ref. 44
as ̇m = ρvA, where ρ is the briny fluid density (ρ = 1,400 kg m−3
),
v is the velocity (v = 8.6 × 10−6
m s−1
) and A is the cross-sectional area of 72 m2
. A
value of 2.0 W m−1
K−1
is used for thermal conductivity42
and 3,400 J kg−1
K−1
is used
for specific heat45
. The value of r∞ refers to the distance between the surface and the
aquifer (that is, 750 m and 4.5 km) and the value used for r0, which represents the
radius of the spring, is 0.5 cm, as suggested in ref. 36
.
Data availability
The authors declare that the data supporting the findings of this study are available
within the article and its Supplementary Information.
References
41. Kirk, R. L. et al. Ultrahigh resolution topographic mapping of Mars with
MRO HiRISE stereo images: Meter scale slopes of candidate Phoenix
landing sites. J. Geophys. Res. 113, E00A24 (2008).
42. Clifford, S. M. A model for the hydrologic and climatic behavior of water
on Mars. J. Geophys. Res. 98, 10973–11016 (1993).
43. Clifford, S. M. et al. Depth of the Martian cryosphere: revised estimates and
implications for the existence and detection of subpermafrost groundwater.
J. Geophys. Res. 115, E07001 (2010).
44. Levy, J. Hydrological characteristics of recurrent slope lineae on Mars:
evidence for liquid flow through regolith and comparisons with Antarctic
terrestrial analogs. Icarus 219, 1–4 (2012).
45. Archer, D. G. & Carter, R. W. Thermodynamic properties of the NaCl H2O
system. 4. Heat capacities of H2O and NaCl (aq) in cold-stable and
supercooled states. J. Phys. Chem. B 104, 8563–8584 (2000).
Nature Geoscience | www.nature.com/naturegeoscience